Case Studies
Industrial and Research Projects with SimVantage

Reactor Optimization · Peer-Reviewed Publication

Reactor Optimization – Sandoz Kundl

Impeller reconfiguration in a 160 m³ bioreactor

Alternating radial and axial impellers made the production fermenter mix faster on less power. The new setup was evaluated in experiments and with CFD (computational fluid dynamics).

• 30%Shorter mixing time
• 20%Lower energy demand
• 3,740MWh saved per year

Scale-Up

From Shake Flask to Stirred Tank

Giving cells in the stirred tank the same flow conditions as in the shake flask

A validated free-surface CFD model calculates the flow conditions cells see in a shaken flask. From there we found the stirred-tank operating points that reproduce them, for NK cells and for three Streptomyces species.

Digital Twin

ECOnti – Digital Twin for Continuous Bioproduction

Mechanistic models and process data for stable continuous operation

Six partners built a digital twin of a continuous E. coli process, from growth through purification. SimVantage added the CFD, so operators saw mixing and shear next to their process data.

• 30Days of stable continuous production

Mixing & Mass Transfer

Mixing & Mass Transfer

Mass transfer and mixing time in an aerated production vessel

A CFD map of mixing time and k_La (oxygen transfer coefficient) showed what limits the process at each operating point. The customer could then raise oxygen transfer where it was short and skip extra agitation where mixing was already good enough.

Hydrodynamics · Peer-Reviewed Publication

Free Surface

Vortex formation in a stirred tank, measured and predicted

Vortex depth, width and volume were measured in a Rushton turbine tank with five baffle configurations and compared with a GPU lattice-Boltzmann free-surface model that needs no fitting factors. A new swirl-number correlation predicts vortex depth for all five.

Scale-Up

Scale-Up with Hybrid Models

CFD-trained models for moving a process between scales

Neural networks trained on CFD results predict shear, oxygen transfer and mixing time in milliseconds, even where correlations fall short. That turns in-silico scale-up into a matter of minutes.